I was reluctant to say it, but I'm not satisfied with Panda's code size. Currently it has 1013 sloc, it should not have been as a simple blog.

I could think of one solution, i.e., the HAppS's approach:

Haskell Datatype -> XML -> XSLT -> CSS -> Javascript

Everything after XML is theme specific, hence outside the scope of Panda. This way I can bring down the amount of code involved.

Unfortunately I might have less time to spend on Panda, so I'll just stfu and code.